Glüer C C, Blake G, Lu Y, Blunt B A, Jergas M, Genant H K
Department of Radiology, University of California, San Francisco 94143-0628, USA.
Osteoporos Int. 1995;5(4):262-70. doi: 10.1007/BF01774016.
Assessment of precision errors in bone mineral densitometry is important for characterization of a technique's ability to detect longitudinal skeletal changes. Short-term and long-term precision errors should be calculated as root-mean-square (RMS) averages of standard deviations of repeated measurements (SD) and standard errors of the estimate of changes in bone density with time (SEE), respectively. Inadequate adjustment for degrees of freedom and use of arithmetic means instead of RMS averages may cause underestimation of true imprecision by up to 41% and 25% (for duplicate measurements), respectively. Calculation of confidence intervals of precision errors based on the number of repeated measurements and the number of subjects assessed serves to characterize limitations of precision error assessments. Provided that precision error are comparable across subjects, examinations with a total of 27 degrees of freedom result in an upper 90% confidence limit of +30% of the mean precision error, a level considered sufficient for characterizing technique imprecision. We recommend three (or four) repeated measurements per individual in a subject group of at least 14 individuals to characterize short-term (or long-term) precision of a technique.
评估骨密度测定中的精密度误差对于表征一项技术检测骨骼纵向变化的能力很重要。短期和长期精密度误差应分别计算为重复测量标准差（SD）的均方根（RMS）平均值和骨密度随时间变化估计值的标准误差（SEE）。对自由度的调整不足以及使用算术平均值而非RMS平均值，可能分别导致对真实不精密度的低估高达41%和25%（对于重复测量）。基于重复测量次数和评估的受试者数量计算精密度误差的置信区间，有助于表征精密度误差评估的局限性。如果各受试者的精密度误差具有可比性，自由度总数为27的检测会得出平均精密度误差的90%上限置信限为+30%，这一水平被认为足以表征技术的不精密度。我们建议在至少14名个体的受试者组中，每人进行三次（或四次）重复测量，以表征一项技术的短期（或长期）精密度。
